CFPB HMDA Resubmission Examination Procedures Guidelines
CFPB Examination Procedures
HMDA Resubmission Guidelines
HMDA Resubmission Schedule and Guidelines These guidelines describe transaction testing that the CFPB will conduct to verify the accuracy of institution-reported HMDA data during HMDA reviews. The guidelines also describe when institutions should be required to correct and resubmit HMDA data.
HMDA Transaction Testing Before starting a HMDA review, CFPB examination teams should select a random sample of entries from the institution’s HMDA Loan Application Register (LAR). CFPB examination teams should also ask the institution to pull the loan or application files that correspond to the HMDA LAR sample entries. Once on-site, CFPB examination teams should check the data in the HMDA LAR sample entries against the loan or application files. The size of the HMDA LAR sample will depend on the size of the institution’s HMDA LAR, as shown in the HMDA Resubmission Schedule (Appendix B). If an institution’s HMDA data are collected through multiple data collection and reporting systems, the CFPB may test samples of entries from each system independently (i.e., select more than one HMDA LAR sample), and may select additional HMDA LAR samples from the system(s) posing the greatest compliance risk. In addition, purchased loans may be sampled separately if they constitute a sufficiently large portion of a covered institution’s HMDA LAR entries. CFPB examination teams will follow these steps when verifying the accuracy of HMDA data: 1) Examination teams will review a HMDA LAR sample and note any errors. The size of the HMDA LAR sample depends on the size of the institution’s HMDA LAR (see Appendix B, HMDA Resubmission Schedule, column A). Examination teams may start the review using the smaller sample shown in Appendix A, Sample Sizes and Thresholds for Streamlined HMDA Review (Streamlined Schedule). If no errors or only a very small number of errors are found during the streamlined review (no more than the number shown in column C of the Streamlined Schedule), no further action is needed and the examination teams may stop the verification procedure. When examination teams identify a number of errors exceeding the number in column C of the Streamlined Schedule, they should always complete a review of the total random sample, as set forth in column B of the HMDA Resubmission Schedule in Appendix B, in order to determine the total number of errors in the full sample and whether the Resubmission Threshold in column C is met. This is necessary — even if errors exceeding the Resubmission Threshold are identified before the review is complete — in order to gauge the full extent of the errors on the overall HMDA LAR and determine any appropriate corrective action.
CFPB
October 2013
Procedures 1
CFPB Examination Procedures
HMDA Resubmission Guidelines
2) If, after reviewing the full HMDA LAR sample, the total number of errors equals or exceeds the number in column C of the HMDA Resubmission Schedule (the Resubmission Threshold), the institution should be required to correct and resubmit its HMDA data.
Resubmission Guidelines I. HMDA Reporters with Fewer than 100,000 Entries on the HMDA LAR
Institutions reporting fewer than 100,000 loans or applications on the HMDA LAR should be required to correct and resubmit HMDA data when 10 percent or more of the HMDA LAR sample entries contain errors. This threshold is set forth in column C of the HMDA Resubmission Schedule, Appendix B below. For example, if a reporting institution’s HMDA LAR contains 1,000 entries, examination teams will review a random sample of 79 files. If there are at least eight HMDA LAR entries with errors (10 percent of the 79 files), the institution should be required to correct and resubmit its entire HMDA LAR. If there are at least four LAR entries with errors within an individual data field (five percent of the 79 files), the institution should be required to correct and resubmit that field. In certain cases, sample error rates below 10 percent — or below five percent in an individual data field — may require resubmission if the errors make analysis of the institution’s data unreliable. II. HMDA Reporters with 100,000 or More Entries on the HMDA LAR
For reporting institutions with 100,000 or more entries on the HMDA LAR, examination teams will review a larger number of entries to ensure that the CFPB has a more precise estimate of the institution’s HMDA LAR error rate. Furthermore, the resubmission standard is more stringent (i.e., the threshold error rate is lower) for these institutions because a lower sample error rate at these institutions could reflect a larger number of errors than a higher sample error rate at an institution with a smaller HMDA LAR. Therefore, institutions with 100,000 or more HMDA LAR entries should be required to correct and resubmit HMDA data when four percent or more of the sample entries contain errors. This threshold is set forth in column C of the HMDA Resubmission Schedule, Appendix B below. For example, if a reporting institution’s HMDA LAR contains 100,000 or more entries, examination teams will review 200 entries; and if there are at least eight entries with errors (four percent of 200 files), the institution should be required to correct and resubmit its entire HMDA LAR. If the error rate in an individual data field is between two and four percent, the institution should be required to correct and resubmit that field. In certain cases, sample error rates below four percent — or below two percent in an individual data field — may require resubmission if the errors make analysis of the institution’s lending unreliable.
CFPB
October 2013
Procedures 2
CFPB Examination Procedures
HMDA Resubmission Guidelines
Illustrative Examples The full HMDA Resubmission Schedule and Sample Sizes and Thresholds for Streamlined HMDA Review are available as Appendices (attached) and are also excerpted below. Sample Sizes and Thresholds for Streamlined HMDA Review Total HMDA LAR entries
Streamlined Review Sample Size
Full review threshold If errors exceed this number, examination teams will continue with review
(A)
(B)
(C)
1,101–99,999
32
1
≥ 100K
90
2
HMDA Resubmission Schedule Total HMDA LAR entries
Total random sample
Resubmission threshold Institutions should be required to resubmit if errors equal or exceed this number
(A)
(B)
(C)
1,101–99,999
79
8
≥ 100K
200
8
Example 1
Institution A has 2,000 entries on its HMDA LAR. The examination team selects a full sample of 79 loans to review. In the first 32 loans reviewed, the examination team does not identify any HMDA LAR entries with errors. The examination team then stops the review, consistent with Appendix A, Samples Sizes and Thresholds for Streamlined HMDA Review. Example 2
Institution B has 200,000 entries on its HMDA LAR. The examination team selects a full sample of 200 loans to review. In the first 90 loans reviewed, the examination team identifies three errors. Therefore the examination team continues to review the additional loans in the sample to determine whether the resubmission threshold is met. The examination team completes the full review of the 200 loans and identifies a total of nine errors. The total number of errors in the sample exceeds the Resubmission Threshold in Appendix B, HMDA Resubmission Schedule, so Institution B should be required to correct and resubmit its HMDA LAR.
CFPB
October 2013
Procedures 3
CFPB Examination Procedures
HMDA Resubmission Guidelines
Appendix A
Sample Sizes and Thresholds for Streamlined HMDA Review If the number of errors in the Streamlined Review Sample is less than or equal to the number in column C below, the review is complete. If the number of errors exceeds the number in column C below, examiners should review the full sample indicated in column B of Appendix B, HMDA Resubmission Schedule, in order to determine whether the Resubmission Threshold is met.
CFPB
Total HMDA LAR entries
Streamlined Review Sample Size
Full review threshold
(A)
(B)
(C)
1–11
Review all
12–20
12
0
21–30
13
0
31–50
15
0
51–70
17
0
71–90
18
0
91–110
28
1
111–130
29
1
131–140
29
1
141–170
29
1
171–190
30
1
191–270
30
1
271–380
30
1
381–750
31
1
751–1,100
31
1
1,101–99,999
32
1
≥ 100K
90
2
October 2013
Procedures 4
CFPB Examination Procedures
HMDA Resubmission Guidelines
Appendix B HMDA Resubmission Schedule1 Institutions should be required to resubmit if the number of errors in the sample equals or exceeds the number in column C.
Total HMDA LAR entries
Total random sample
Resubmission threshold
(A)
(B)
(C)
1–11
All
12–20
All
2
21–30
All
2
31–50
28
3
51–70
29
3
71–90
38
4
91–110
39
4
111–130
47
5
131–140
49
5
141–170
56
6
171–190
57
6
191–270
59
6
271–380
68
7
381–750
69
7
751–1,100
79
8
1,101–99,999
79
8
≥ 100K
200
8
1
The HMDA Resubmission Schedule and the Streamlined HMDA Review Schedule apply to both depository and nondepository HMDA reporting institutions.
CFPB
October 2013
Procedures 5
HMDA Resubmission Guidelines
HMDA Resubmission Schedule and Guidelines These guidelines describe transaction testing that the CFPB will conduct to verify the accuracy of institution-reported HMDA data during HMDA reviews. The guidelines also describe when institutions should be required to correct and resubmit HMDA data.
HMDA Transaction Testing Before starting a HMDA review, CFPB examination teams should select a random sample of entries from the institution’s HMDA Loan Application Register (LAR). CFPB examination teams should also ask the institution to pull the loan or application files that correspond to the HMDA LAR sample entries. Once on-site, CFPB examination teams should check the data in the HMDA LAR sample entries against the loan or application files. The size of the HMDA LAR sample will depend on the size of the institution’s HMDA LAR, as shown in the HMDA Resubmission Schedule (Appendix B). If an institution’s HMDA data are collected through multiple data collection and reporting systems, the CFPB may test samples of entries from each system independently (i.e., select more than one HMDA LAR sample), and may select additional HMDA LAR samples from the system(s) posing the greatest compliance risk. In addition, purchased loans may be sampled separately if they constitute a sufficiently large portion of a covered institution’s HMDA LAR entries. CFPB examination teams will follow these steps when verifying the accuracy of HMDA data: 1) Examination teams will review a HMDA LAR sample and note any errors. The size of the HMDA LAR sample depends on the size of the institution’s HMDA LAR (see Appendix B, HMDA Resubmission Schedule, column A). Examination teams may start the review using the smaller sample shown in Appendix A, Sample Sizes and Thresholds for Streamlined HMDA Review (Streamlined Schedule). If no errors or only a very small number of errors are found during the streamlined review (no more than the number shown in column C of the Streamlined Schedule), no further action is needed and the examination teams may stop the verification procedure. When examination teams identify a number of errors exceeding the number in column C of the Streamlined Schedule, they should always complete a review of the total random sample, as set forth in column B of the HMDA Resubmission Schedule in Appendix B, in order to determine the total number of errors in the full sample and whether the Resubmission Threshold in column C is met. This is necessary — even if errors exceeding the Resubmission Threshold are identified before the review is complete — in order to gauge the full extent of the errors on the overall HMDA LAR and determine any appropriate corrective action.
CFPB
October 2013
Procedures 1
CFPB Examination Procedures
HMDA Resubmission Guidelines
2) If, after reviewing the full HMDA LAR sample, the total number of errors equals or exceeds the number in column C of the HMDA Resubmission Schedule (the Resubmission Threshold), the institution should be required to correct and resubmit its HMDA data.
Resubmission Guidelines I. HMDA Reporters with Fewer than 100,000 Entries on the HMDA LAR
Institutions reporting fewer than 100,000 loans or applications on the HMDA LAR should be required to correct and resubmit HMDA data when 10 percent or more of the HMDA LAR sample entries contain errors. This threshold is set forth in column C of the HMDA Resubmission Schedule, Appendix B below. For example, if a reporting institution’s HMDA LAR contains 1,000 entries, examination teams will review a random sample of 79 files. If there are at least eight HMDA LAR entries with errors (10 percent of the 79 files), the institution should be required to correct and resubmit its entire HMDA LAR. If there are at least four LAR entries with errors within an individual data field (five percent of the 79 files), the institution should be required to correct and resubmit that field. In certain cases, sample error rates below 10 percent — or below five percent in an individual data field — may require resubmission if the errors make analysis of the institution’s data unreliable. II. HMDA Reporters with 100,000 or More Entries on the HMDA LAR
For reporting institutions with 100,000 or more entries on the HMDA LAR, examination teams will review a larger number of entries to ensure that the CFPB has a more precise estimate of the institution’s HMDA LAR error rate. Furthermore, the resubmission standard is more stringent (i.e., the threshold error rate is lower) for these institutions because a lower sample error rate at these institutions could reflect a larger number of errors than a higher sample error rate at an institution with a smaller HMDA LAR. Therefore, institutions with 100,000 or more HMDA LAR entries should be required to correct and resubmit HMDA data when four percent or more of the sample entries contain errors. This threshold is set forth in column C of the HMDA Resubmission Schedule, Appendix B below. For example, if a reporting institution’s HMDA LAR contains 100,000 or more entries, examination teams will review 200 entries; and if there are at least eight entries with errors (four percent of 200 files), the institution should be required to correct and resubmit its entire HMDA LAR. If the error rate in an individual data field is between two and four percent, the institution should be required to correct and resubmit that field. In certain cases, sample error rates below four percent — or below two percent in an individual data field — may require resubmission if the errors make analysis of the institution’s lending unreliable.
CFPB
October 2013
Procedures 2
CFPB Examination Procedures
HMDA Resubmission Guidelines
Illustrative Examples The full HMDA Resubmission Schedule and Sample Sizes and Thresholds for Streamlined HMDA Review are available as Appendices (attached) and are also excerpted below. Sample Sizes and Thresholds for Streamlined HMDA Review Total HMDA LAR entries
Streamlined Review Sample Size
Full review threshold If errors exceed this number, examination teams will continue with review
(A)
(B)
(C)
1,101–99,999
32
1
≥ 100K
90
2
HMDA Resubmission Schedule Total HMDA LAR entries
Total random sample
Resubmission threshold Institutions should be required to resubmit if errors equal or exceed this number
(A)
(B)
(C)
1,101–99,999
79
8
≥ 100K
200
8
Example 1
Institution A has 2,000 entries on its HMDA LAR. The examination team selects a full sample of 79 loans to review. In the first 32 loans reviewed, the examination team does not identify any HMDA LAR entries with errors. The examination team then stops the review, consistent with Appendix A, Samples Sizes and Thresholds for Streamlined HMDA Review. Example 2
Institution B has 200,000 entries on its HMDA LAR. The examination team selects a full sample of 200 loans to review. In the first 90 loans reviewed, the examination team identifies three errors. Therefore the examination team continues to review the additional loans in the sample to determine whether the resubmission threshold is met. The examination team completes the full review of the 200 loans and identifies a total of nine errors. The total number of errors in the sample exceeds the Resubmission Threshold in Appendix B, HMDA Resubmission Schedule, so Institution B should be required to correct and resubmit its HMDA LAR.
CFPB
October 2013
Procedures 3
CFPB Examination Procedures
HMDA Resubmission Guidelines
Appendix A
Sample Sizes and Thresholds for Streamlined HMDA Review If the number of errors in the Streamlined Review Sample is less than or equal to the number in column C below, the review is complete. If the number of errors exceeds the number in column C below, examiners should review the full sample indicated in column B of Appendix B, HMDA Resubmission Schedule, in order to determine whether the Resubmission Threshold is met.
CFPB
Total HMDA LAR entries
Streamlined Review Sample Size
Full review threshold
(A)
(B)
(C)
1–11
Review all
12–20
12
0
21–30
13
0
31–50
15
0
51–70
17
0
71–90
18
0
91–110
28
1
111–130
29
1
131–140
29
1
141–170
29
1
171–190
30
1
191–270
30
1
271–380
30
1
381–750
31
1
751–1,100
31
1
1,101–99,999
32
1
≥ 100K
90
2
October 2013
Procedures 4
CFPB Examination Procedures
HMDA Resubmission Guidelines
Appendix B HMDA Resubmission Schedule1 Institutions should be required to resubmit if the number of errors in the sample equals or exceeds the number in column C.
Total HMDA LAR entries
Total random sample
Resubmission threshold
(A)
(B)
(C)
1–11
All
12–20
All
2
21–30
All
2
31–50
28
3
51–70
29
3
71–90
38
4
91–110
39
4
111–130
47
5
131–140
49
5
141–170
56
6
171–190
57
6
191–270
59
6
271–380
68
7
381–750
69
7
751–1,100
79
8
1,101–99,999
79
8
≥ 100K
200
8
1
The HMDA Resubmission Schedule and the Streamlined HMDA Review Schedule apply to both depository and nondepository HMDA reporting institutions.
CFPB
October 2013
Procedures 5
